Marvell fires first shot in ARM-x64 server scrap

Armada XP cruisers versus Xeon battleships


Chip maker Marvell, which acquired the Xscale ARM RISC chip business from Intel in 2006, is now officially gunning for the chip giant in the market for low-powered, cloudy infrastructure servers. If there is such a market.

At the ARM Techcon 2010 in Intel's hometown of Santa Clara, California, Marvell unveiled its quad-core Armada XP processor, which it claims offers up to five times the performance per watt as an equivalent (but unspecified) server chip from Intel.

The Armada XP chip is based on the ARM v7 MP core created by Marvell, which is a licensee of ARM Holding's various Cortex ARM processor designs. The 32-bit Armada XP is implemented in 40 nanometer wafer baking processes and has four cores running at 1.6 GHz with a shared 2 MB L2 cache memory for data to play in.

The chip will include variants that support 64-bit DDR2, DDR3, and DDR3 low-voltage memory chips. For on-chip DDR3 controllers, the memory can run at to 800 MHz and the memory controller on the Armada XP chip has ECC memory scrubbing. Marvell has not divulged how much L1 data and instruction cache is on each of the four cores on the Armada XP processor.

The Armada XP chip has a whole bunch of system components integrated onto its silicon, which will make it popular among puny system providers (that referred to the servers as well as the companies making the boxes). The chip includes four PCI-Express 2.0 x4 interfaces and four Gigabit Ethernet controllers etched into its silicon; it has 16 SERDES lanes for implementing USB, PCI-Express, SATA, SGMII, and QSGMII ports on the mobo.

Marvell Armada XP ARM chip

The Armada XP ARM processor from Marvell

Marvell says that the Armada XP chip can support symmetric multiprocessing (where all four cores share the main memory) as well as asymmetric multiprocessing (where the cores and their memory are separated) as well as a mixed mode using what it calls "hardware-based cache coherence."

Operating at 1.6 GHz, the four cores collectively deliver 16,600 Dhrystone MIPS (DMIPS) of performance at under 10 watts of power dissipation. It is hard to say how this will stack up to Intel's low-powered Atom and high-powered Xeon processors for server workloads, but the Armada XP is certainly in the performance range that makes it an interesting contender for simple server stacks and workloads that don't need more than 4 GB of main memory. (For an interesting performance comparison of ARM, MIPS, and Atom chips, check this out.)

What we really need to settle this performance matchup between ARM, Atom, and Xeon is a LAMP-stack bake-off with a real workload or ten. Hopefully when Marvell starts shipping the Armada XPs - and partners start building servers using them - they will come fully armed with such data.

Marvell is starting to sample the Armada XP chips now, and says that it expects uptake among makers of Web servers, network equipment, and home NAS and media servers. In addition to performance of these chips, everyone wants to know what it will cost to slap together an entry server.

That's something that Smooth-Stone, a startup that is still in stealth mode that is working on ARM-based servers, will hopefully shed some light on soon. The server business needs a little more competition to keep it honest. ®

Broader topics


Other stories you might like

  • DuckDuckGo tries to explain why its browsers won't block some Microsoft web trackers
    Meanwhile, Tails 5.0 users told to stop what they're doing over Firefox flaw

    DuckDuckGo promises privacy to users of its Android, iOS browsers, and macOS browsers – yet it allows certain data to flow from third-party websites to Microsoft-owned services.

    Security researcher Zach Edwards recently conducted an audit of DuckDuckGo's mobile browsers and found that, contrary to expectations, they do not block Meta's Workplace domain, for example, from sending information to Microsoft's Bing and LinkedIn domains.

    Specifically, DuckDuckGo's software didn't stop Microsoft's trackers on the Workplace page from blabbing information about the user to Bing and LinkedIn for tailored advertising purposes. Other trackers, such as Google's, are blocked.

    Continue reading
  • Despite 'key' partnership with AWS, Meta taps up Microsoft Azure for AI work
    Someone got Zuck'd

    Meta’s AI business unit set up shop in Microsoft Azure this week and announced a strategic partnership it says will advance PyTorch development on the public cloud.

    The deal [PDF] will see Mark Zuckerberg’s umbrella company deploy machine-learning workloads on thousands of Nvidia GPUs running in Azure. While a win for Microsoft, the partnership calls in to question just how strong Meta’s commitment to Amazon Web Services (AWS) really is.

    Back in those long-gone days of December, Meta named AWS as its “key long-term strategic cloud provider." As part of that, Meta promised that if it bought any companies that used AWS, it would continue to support their use of Amazon's cloud, rather than force them off into its own private datacenters. The pact also included a vow to expand Meta’s consumption of Amazon’s cloud-based compute, storage, database, and security services.

    Continue reading
  • Atos pushes out HPC cloud services based on Nimbix tech
    Moore's Law got you down? Throw everything at the problem! Quantum, AI, cloud...

    IT services biz Atos has introduced a suite of cloud-based high-performance computing (HPC) services, based around technology gained from its purchase of cloud provider Nimbix last year.

    The Nimbix Supercomputing Suite is described by Atos as a set of flexible and secure HPC solutions available as a service. It includes access to HPC, AI, and quantum computing resources, according to the services company.

    In addition to the existing Nimbix HPC products, the updated portfolio includes a new federated supercomputing-as-a-service platform and a dedicated bare-metal service based on Atos BullSequana supercomputer hardware.

    Continue reading

Biting the hand that feeds IT © 1998–2022